Prelander ladders: five angles, one survivor

A pop visitor is an interruption you purchased. They were reading something else; your page appeared. The prelander's job is to convert that accident into intent inside about two seconds of attention, and no single page design earns that reliably across offers and GEOs. So we never test one prelander. We test five, as a ladder, under rules strict enough that the result means something.

The five-angle structure

Every ladder ships the same skeleton of angles: a news-style advertorial, a quiz or picker, a scarcity or urgency frame, a plain benefits page, and one wildcard that breaks the pattern on purpose. The wildcard exists because convention is a local maximum: in the subscription split, a deliberately sparse one-paragraph page beat the advertorial that eleven years of habit said would win, and took ROI from 31% to 118%. Wildcards win roughly one ladder in five. That rate pays for every ladder they lose.

Judge at the payout event, never the click

Each angle gets equal rotation in the tracker and a minimum spend of three times the offer payout before any verdict, the same 3× logic the zone rules use. The judging metric is verified conversions per angle from S2S postbacks, full stop. CTR is recorded for diagnosis and excluded from verdicts: quiz pages routinely produce the best click-through and mediocre paid conversion, because a quiz can manufacture clicks from people who were never going to buy anything.

The 1.5-second budget

Tier 2–3 pop traffic is mostly mid-range Android on cellular, so every prelander must be interactive in under 1.5 seconds on that hardware. In practice that forces: static HTML, critical CSS inlined, total images under 60KB, system fonts or one subsetted file, and no third-party scripts except the tracker's. Test on a real device, not your desktop's throttled tab. A page that misses the budget does not enter the ladder; speed is a precondition, not an optimization.

Scaling the survivor

The winner takes 100% of rotation, and only then does the media side move: bids climb inside the existing whitelist first, new zones second, new GEOs last. Changing the page and the traffic at the same time destroys attribution, so ladders always run against a frozen media configuration. Expect the survivor to fatigue; when its conversion rate decays about 20% from peak, the next ladder starts, seeded with the survivor's skeleton and four new challengers.

Ladder rules on one card

1. Five angles: advertorial, quiz, urgency, plain benefits, wildcard.
2. Equal rotation; 3× payout minimum spend per angle.
3. Verdict = verified payout events per angle. CTR never decides.
4. Under 1.5s interactive on mid-range Android or it does not run.
5. Media config frozen for the ladder's whole life.
6. Rerun at ~20% conversion decay from the winner's peak.